• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권3호
• /
• pp.911-931
• /
• 2021

Recently, most cloud services use Docker container environment to provide their services. However, there are no researches to evaluate the performance of communication libraries for multi-GPU based distributed deep learning in a Docker container environment. In this paper, we propose an efficient communication architecture for multi-GPU based deep learning in a Docker container environment by evaluating the performances of various communication libraries. We compare the performances of the parameter server architecture and the All-reduce architecture, which are typical distributed deep learning architectures. Further, we analyze the performances of two separate multi-GPU resource allocation policies - allocating a single GPU to each Docker container and allocating multiple GPUs to each Docker container. We also experiment with the scalability of collective communication by increasing the number of GPUs from one to four. Through experiments, we compare OpenMPI and MPICH, which are representative open source MPI libraries, and NCCL, which is NVIDIA's collective communication library for the multi-GPU setting. In the parameter server architecture, we show that using CUDA-aware OpenMPI with multi-GPU per Docker container environment reduces communication latency by up to 75%. Also, we show that using NCCL in All-reduce architecture reduces communication latency by up to 93% compared to other libraries.

• The Journal of Asian Finance, Economics and Business
• /
• 제10권2호
• /
• pp.1-14
• /
• 2023

The purpose of this study is to discover the possible solutions for the problem faced by academicians during online learning by means of employing qualitative and quantitative approaches. Using a qualitative approach, selected academicians were interviewed, and their feedback was transcribed and used to develop the survey instrument. The quantitative research design was later used to determine the most plausible solutions for the problem that could be obtained and implemented by distributing the questionnaire to academicians at a public university. Multiple regression analysis results indicate that work-life conflict and lack of support are the main contributors to academicians' mental and emotional health issues. The study's major findings help higher education institutions craft appropriate strategies to enhance the effectiveness of online teaching and learning by providing the necessary support to the academicians. The study's findings suggest that academicians should separate work and family requirements to concentrate on their job. Furthermore, the immediate supervisor must be considerate in determining the number of tasks, the deadlines, and the assistance required to complete the task. Lastly, academicians must equip themselves with emotional intelligence to cope with stressors.

• 한국산학기술학회논문지
• /
• 제17권3호
• /
• pp.668-676
• /
• 2016

본 연구는 국내 중학교 건축 계획 시 참조점을 구하기 위해 관련 연구가 부족했던 일본의 중학교 건축을 대상으로 그 계획특성을 살펴보았다. 이를 위해 1990년대 이후 준공된 14개 학교들을 선별해 이들을 대상으로 교육 및 생활공간, 기타 주요 공간 등의 영역별 계획특성을 분석하였다. 분석대상 학교들의 계획특성을 정리하면 다음과 같다. 1) 분석대상 학교들은 학교 운영방식에 따라 교과교실형 학교와 특별교실형 학교로 크게 구분할 수 있는데 이에 따라 가장 기본적인 공간이 되는 교육 및 생활공간의 계획특성에서 차이가 나타난다. 2) 교과교실형 학교는 교과교실, 홈베이스, 미디어 스페이스, 교사공간 등의 요소로 교육 및 생활공간이 구성되며 홈베이스와 교과교실과의 인접여부에 따라 교실 부속형과 독립형의 두 가지 유형으로 크게 구분된다. 3) 특별교실형 학교는 학급교실과 다목적 공간으로 교육 및 생활공간이 구성되는데, 다목적 공간과 학급교실이 직접 연결되어 있는 유형과 분리되어 있는 두 가지 유형으로 구분된다. 4) 특별교실은 지역개방과 과목별 특성을 강조할 수 있는 계획특성이 나타나며 부속 공간들의 연계로 공간의 분화특성이 나타난다. 5) 도서실은 개별실이 아닌 컴퓨터실, 시청각실, 다목적 홀 등과 연계된 복합구성으로 계획되며 벽으로 구획하지 않고 개방적으로 구성하는 경향이 강하다. 6) 체육시설로는 옥내 체육관이 기본적으로 설치되며 이와 함께 무도장, 옥외 수영장 등의 설치 빈도가 높다. 이들은 체육수업뿐만 아니라 방과 후 활동을 지원하기 위해 설치된 측면도 크다. 7) 테라스, 발코니, 외부계단 등의 적극적인 계획을 통해 외부공간을 입체적으로 확보하고 있으며 이들이 다양한 기능을 가지도록 계획하고 있다.

• The Korean Journal of Ecology
• /
• 제23권2호
• /
• pp.89-100
• /
• 2000

On community data. sampled in regular intervals on a long-term basis. artificial neural networks were implemented to extract information on characterizing patterns of community changes. The Adaptive Resonance Theory and Kohonen Network were both utilized in learning benthic macroinvertebrate communities in the Soktae Stream of the Suyong River collected monthly for three years. Initially, by regarding each monthly collection as a separate sample unit, communities were grouped into similar patterns after training with the networks. Subsequently, changes in communities in a sequence of samplings (e.g., two-month, four-month, etc.) were given as input to the networks. After training, it was possible to recognize new data set in line with the sampling procedure. Through the comparative study on benthic macroinvertebrates with these learning processes, patterns of community changes in chironomids diverged while those of the total benthic macro-invertebrates tended to be more stable.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제51권2호
• /
• pp.69-77
• /
• 2002

Speaker verification can be classified in two categories, text-dependent speaker verification and text-independent speaker verification. In this paper, we discuss text-dependent speaker verification. Text-dependent speaker verification system determines whether the sound characteristics of the speaker are equal to those of the specific person or not. In this paper we obtain the speaker data using a sound card in various noisy conditions, apply a new Hidden LMS (Least Mean Square) adaptive algorithm to it, and extract LPC (Linear Predictive Coding)-cepstrum coefficients as feature vectors. Finally, we use a competitive learning neural network for speaker verification. The proposed hidden LMS adaptive filter using a neural network reduces noise and enhances features in various noisy conditions. We construct a separate neural network for each speaker, which makes it unnecessary to train the whole network for a new added speaker and makes the system expansion easy. We experimentally prove that the proposed method improves the speaker verification performance.

• ETRI Journal
• /
• 제45권5호
• /
• pp.862-873
• /
• 2023

Determining whether an autonomous self-driving agent is in the middle of an intersection can be extremely difficult when relying on visual input taken from a single camera. In such a problem setting, a wider range of views is essential, which drives us to use three cameras positioned in the front, left, and right of an agent for better intersection recognition. However, collecting adequate training data with three cameras poses several practical difficulties; hence, we propose using data collected from one camera to train a three-camera model, which would enable us to more easily compile a variety of training data to endow our model with improved generalizability. In this work, we provide three separate fusion methods (feature, early, and late) of combining the information from three cameras. Extensive pedestrian-view intersection classification experiments show that our feature fusion model provides an area under the curve and F1-score of 82.00 and 46.48, respectively, which considerably outperforms contemporary three- and one-camera models.

• International Journal of Computer Science & Network Security
• /
• 제22권8호
• /
• pp.260-268
• /
• 2022

The article considers the training, competitiveness of specialists, professional mobility, professionalism and competence of specialists in the context of distance learning. The advantages of distance learning are shown. The characteristic features of distance learning in the preparation of students and in the implementation of these technologies in the educational process of higher educational institutions are determined. Competitiveness, professional mobility, professionalism and competence of a specialist are qualities that determine a person's life and work success. Professional mobility is interpreted as a systemic quality of a specialist's personality, which includes a whole range of knowledge, skills, abilities, personal qualities, value orientations, and so on. The vision of mobility of specialists by foreign scientists is presented. It is noted that the classification of professional mobility presented in the article makes it possible to organize various movements from a single position, to present them as separate manifestations of the general process of professional and pedagogical mobility, to determine which type of mobility ensures the performance of certain social functions. It was found that mobility can be differentiated into differentiated and intergeneration. According to the subject, individual and group mobility are distinguished; according to the direction - internal and external. The classification of employees according to their attitude to mobility is shown, which can be divided into the following groups: actually mobile; potentially mobile; actually stable; potentially stable.

• 대한음성학회지:말소리
• /
• 제56호
• /
• pp.29-47
• /
• 2005

Investigating and comparing English vowels $/{\varepsilon}/\;and\;/{\ae}/$, and their-supposedly- corresponding vowels in Korean /ㅔ/ and /ㅐ/, this study addresses the following questions: Do Koreans pronounce/ㅔ/ and /ㅐ/ differently? Do they pronounce English $/{\varepsilon}/\;and\;/{\ae}/$ differently? And what is the relationship between the Korean vowels and the English vowels? Is the conventional correspondence (/ㅔ:${\varepsilon}/$, and /ㅐ/:${\ae)/$ appropriate? The results showed that 24 Korean male college students distinguish neither Korean /ㅔ/ and /ㅐ/ nor English $/{\varepsilon}/\;and\;/{\ae}/$, thus suggesting that their inability to distinguish the two vowels in their native tongue has an effect on their production of the English vowels. While not distinguishable within a language, Korean /ㅔ/ and /ㅐ/ still form a separate group from English $/{\varepsilon}/\;and\;/{\ae}/$. But Korean-Produced $/{\varepsilon}/\;and\;/{\ae}/$ are significantly different from American-produced $/{\varepsilon}/\;and\;/{\ae}/:$ Korean-produced $/{\varepsilon}/\;and\;/{\ae}/$ have much lower F1 and F2 than American-produced counterparts. Accordingly it is advised that, in learning English pronunciation, Korean students should be instructed to take the English vowel system as a separate system from Korean one, and thus, not to substitute Korean vowels for English vowels. And students should be provided with specific instructions on the articulatory differences between English vowels and Korean vowels. For example, Cey should be instructed to lower their jaws more for English $/{\varepsilon}/\;and\;/{\ae}/$ than for Korean /ㅔ/ and /ㅐ/.

• Nutrition Research and Practice
• /
• 제5권4호
• /
• pp.294-300
• /
• 2011

Takju is a Korean alcoholic beverage made from rice, and is brewed with the yeast Saccharomyces cerevisiae. This study was conducted to evaluate the effects of exercise training and moderate Takju consumption on learning ability in 6-week old Sprague-Dawley male rats. The rats were treated with exercise and alcohol for 4 weeks in six separate groups as follows: non-exercised control (CC), exercised control (EC), non-exercised consuming ethanol (CA), exercised consuming ethanol (EA), non-exercised consuming Takju (CT), and exercised consuming Takju (ET). An AIN-93M diet was provided ad libitum. Exercise training was performed at a speed of 10 m/min for 15 minutes per day. Ethanol and Takju were administered daily for 6-7 hours to achieve an intake of about 10 ml after 12 hours of deprivation, and, thereafter, the animals were allowed free access to deionized water. A Y-shaped water maze was used from the third week to understand the effects of exercise and alcohol consumption on learning and memory. After sacrifice, brain acetylcholinesterase (AChE) activity was analyzed. Total caloric intake and body weight changes during the experiment were not significantly different among the groups. AChE activity was not significantly different among the groups. The number of errors for position reversal training in the maze was significantly smaller in the EA group than that in the CA and ET groups, and latency times were shorter in the EA group than those in the CC, EC, CT, and ET groups. The latency difference from the first to the fifth day was shortest in the ET group. The exercised groups showed more errors and latency than those of the non-exercised groups on the first day, but the data became equivalent from the second day. The results indicate that moderate exercise can increase memory and learning and that the combination of exercise and Takju ingestion may enhance learning ability.

• 전기전자학회논문지
• /
• 제24권2호
• /
• pp.475-485
• /
• 2020

최근, 별도의 센서를 설치하지 않고 센서가 포함된 사용자의 기기로부터 제공되는 실시간 센싱 데이터를 가지고 새로운 센싱 서비스를 제공하는 크라우드센싱(Crowdsensing) 시스템이 주목받고 있다. 크라우드센싱 시스템에서는 사용자의 조작실수나 통신 문제로 인해 의미 없는 데이터가 제공되거나 보상을 얻기 위해 거짓 데이터를 제공할 수 있어 해당 이상 데이터의 탐지 및 제거가 크라우드센싱 서비스의 질을 결정짓는다. 이러한 이상데이터를 탐지하기 위해 제안되었던 방법들은 크라우드센싱의 빠른 변화 환경에 효율적이지 않다. 본 논문은 머신러닝 기술을 활용하여 지속적이고 빠르게 변화하는 센싱 데이터의 특징을 추출하고 적절한 알고리즘을 통해 모델링하여 이상데이터를 탐지하는 방법을 제안한다. 지도학습의 딥러닝 이진 분류 모델과 비지도학습의 오토인코더 모델을 사용하여 제안 시스템의 성능 및 실현 가능성을 보인다.